We have an ECS container running on an EC2 backed instance. Assuming we don't do any cluster maintenance that impacts the autoscaling group, the number of containers using the group, the image type backing the group, the parameters of containers that use the group, etc...how likely is it that AWS will restart that container without warning? We realize that Fargate can yank and redeploy containers at any time, but EC2 instances created for the purpose of running and scaling containers should be stable unless user-initiated maintenance is performed, correct?
After reading this article, I'm pretty confident that my container will remain running until and unless we do maintenance on the cluster that would disturb the load on the cluster. https://repost.aws/knowledge-center/ecs-task-stopped. We have reason to be paranoid that our task will restart unexpectedly. The timing of restarts, if necessary is critical. Specifically we don't want it to restart within a particular time window. Is there any more of a chance that this will restart than say a typical EC2 VM running this process? We will continue to make use of EC2 instead of ECS with EC2 backing if the odds of the container restarting is greater based on AWS initiated maintenance, so any thoughts are appreciated.
AWS OFFICIALUpdated 2 years ago
